Why 627 BindCraft2 trajectories produced zero binders
- 1Shenzhen UniBindMD Computational Biotechnology Co., Ltd.
Abstract
De novo protein-binder design with deep-learning pipelines such as BindCraft2 has been reported to deliver experimental success rates of 10-100% against well-structured targets. Here we report the opposite tail: two BindCraft2 (v1.0.3) cyclic-peptide campaigns targeting the shallow K81 lactylation epitope of human lactate dehydrogenase A (LDHA) produced zero accepted designs from 627 trajectories. Retrospective analysis of the campaign artifacts shows that 97% of trajectories were rejected at the first ("screen") stage by a single confidence gate (binder pLDDT < 0.60), not because the interface failed to form: the interface metric i_pTM reached 0.69 and the target was confidently folded (pTM = 0.90). Critically, the mid-project decision to disable the "lysine trick" (forced_targeting) was backwards -- that feature moved 17 trajectories past the screen versus 1 without it, and raised interface confidence. We document two preset/metric mismatches that set a macrocycle campaign up to fail regardless of budget: a screen-stage pLDDT gate calibrated for globular binders, and a binder_contacts loss that is degenerate for short peptides. We describe the exact artifacts BindCraft2 writes and distill them into a reusable post-mortem protocol, and we include a complete how-to-use-BindCraft2 guide so the failure analysis also serves as a tutorial.
Related
- De Novo Design and AlphaFold3 Evaluation of Protein Binders Targeting Specific Sites of MAP4K4
- Tag-stabilized fluorescent protein binders for endogenous cell type-specific protein labelling
- BAGEL-CAR: Reflections on the Bits to Binders Competition
- De novo design of mini-proteins as specific binders to PACAP-38, a key mediator of migraine
- Can Folding Models Tell Binders from Bluffers? Evidence from POISK: The Patent-Derived Antibody Dataset
BibTeX
@misc{fu2026bindcraft2,
title = {{Why 627 BindCraft2 trajectories produced zero binders}},
author = {W. Fu and J. Chen and X. Wu},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.13.758722},
url = {https://www.biorxiv.org/content/10.64898/2026.10.13.758722v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Why 627 BindCraft2 trajectories produced zero binders
AU - W. Fu
AU - J. Chen
AU - X. Wu
PY - 2026
DA - 2026/10/13
PB - bioRxiv
DO - 10.64898/2026.10.13.758722
UR - https://www.biorxiv.org/content/10.64898/2026.10.13.758722v1
AB - De novo protein-binder design with deep-learning pipelines such as BindCraft2 has been reported to deliver experimental success rates of 10-100% against well-structured targets. Here we report the opposite tail: two BindCraft2 (v1.0.3) cyclic-peptide campaigns targeting the shallow K81 lactylation epitope of human lactate dehydrogenase A (LDHA) produced zero accepted designs from 627 trajectories. Retrospective analysis of the campaign artifacts shows that 97% of trajectories were rejected at the first ("screen") stage by a single confidence gate (binder pLDDT < 0.60), not because the interface failed to form: the interface metric i_pTM reached 0.69 and the target was confidently folded (pTM = 0.90). Critically, the mid-project decision to disable the "lysine trick" (forced_targeting) was backwards -- that feature moved 17 trajectories past the screen versus 1 without it, and raised interface confidence. We document two preset/metric mismatches that set a macrocycle campaign up to fail regardless of budget: a screen-stage pLDDT gate calibrated for globular binders, and a binder_contacts loss that is degenerate for short peptides. We describe the exact artifacts BindCraft2 writes and distill them into a reusable post-mortem protocol, and we include a complete how-to-use-BindCraft2 guide so the failure analysis also serves as a tutorial.
ER -